This section describes the techniques Berann used to manipulate landscapes. To most readers the process of viewing a Berann panorama seems like a random series of happy discoveries. Your eyes move from one feature to another, lingering occasionally in places that attract your interest, and then, perhaps, your focus widens to regard the sweep of the entire landscape. However, your visual journey through a Berann panorama is not entirely a matter of choice and chance. The physical structure of the landscape has been altered to create pathways that subtly guide you to selected features of significance.
Many of Berann’s techniques for manipulating landscapes were unorthodox. He often would take questionable liberties (at least from a cartographic perspective) with geographic reality for the convenience of telling a panorama’s story. Whether or not you agree with his techniques, knowing what can be done is valuable for understanding the problems associated with 3D landscape presentation in general.
Perspective
Perspective provides the framework for building a panorama and governs how much of the world the viewer will see. (Figure 6) In general, Berann used more perspective on small-scale scenes than on large-scale scenes. Increasing the amount of perspective increases the field of view, thereby compressing more background information into the finite width of the panorama. Use of perspective enhances realism, unless applied excessively, in which case it tends to pinch background areas unnaturally. Large-scale panoramas employ very little perspective, for the purpose of moving background features closer to the viewer (Garfield, 1992). For example, Berann minimized perspective on the Yosemite panorama to accentuate the lofty peaks along the Sierra Nevada crest, which would be indistinguishable if greater amounts of perspective were used. He compensated for the lack of optical perspective by exaggerating other graphical cues—cloud formations converge toward a false vanishing point, and the differing foreground and background colors enhance aerial perspective. On the North Cascades panorama, on the other hand, Berann used more perspective to pull peripheral landmarks, such as Mt. Rainier, the city of Seattle, and Vancouver Island, into the scene.
Orientation
The cartographic convention of north orientation is not a major factor for determining the orientation (or viewing direction) of large-scale panoramas. None of Berann’s NPS panoramas is oriented due north: North Cascades is oriented to the southwest, Yosemite to the east, Yellowstone to the south, and Denali to the northwest. Berann’s goals when selecting orientation were to maximize the visibility of important local features and to show the broader geographic context. When depicting alpine mountains in the northern hemisphere, such as the Alps or North Cascades, he often used southwest orientation. Looking southwest reveals northeast mountain faces, which tend to be steeper, more distinctive, and more glaciated. Prevailing southwest winds transport summit snows to these lee slopes, which also get less direct sunlight.
Berann preferred the view from lowlands toward highlands. In fact, most of his panoramas have a lofty mountain range as the backdrop. On large scale panoramas, viewers tend to become disoriented when looking from highlands toward lowlands, despite our real-world familiarity with downhill vistas from mountain peaks. This phenomenon received prominent attention during the controversial court martial of the U.S. pilot who clipped the cable of a ski gondola in the Italian Alps in 1998, tragically killing 20 people. The legal defense team for the pilot presented a digital fly-through animation, which was shown on CBS national television news in the U.S., to demonstrate the difficulties of judging elevation when traveling downhill through a narrow mountain valley (Visual Forensics, 1999).
Berann sometimes added twist (skewed orientation) to his scenes to show topographic features with cultural importance that otherwise would be outside the field of view. This is evident in two NPS panoramas: Yosemite was skewed to the north to include Hetch Hetchy Reservoir, the site of a landmark conservation-versus-development battle in the early nineteen hundreds. North Cascades was skewed to the south to include Mt. Rainier—the defining landmark for the millions of people who live in Seattle and the Puget Sound lowlands (see Appendix A for illustration).
Projection plane
Cartographically speaking, Berann did not accept that the Earth is flat, even on large-scale panoramas. He added curvature to the projection plane (the theoretical flat base upon which 3D terrain projects upwards) to enhance viewing. On a typical 3D scene with a low-elevation view, the horizon is visible, and the landscape looks realistic, but tall features in the foreground obscure the background. Conversely, high elevation views show background terrain better, but, without the horizon, they look too much like conventional maps.
To solve this problem Berann emulated the view as seen from an airplane. (Figure 7) From high above the Earth the horizon is always visible, yet when you shift your eyes downward the view gradually becomes less oblique and more planimetric. To bring this effect to a panorama, Berann tilted the projection plane toward the viewer and, from a point about two thirds of the way into the scene, added convex curvature to flatten the horizon. The end result is a panorama that combines the best of both worlds: the foreground and middleground (where the important information resides) appear map-like while the background appears realistic, complete with a horizon and sky (Patterson, 1999).
Berann’s manipulation of the projection plane is most evident on small scale panoramas, such as his 1986 view of Germany, in which the relatively low relief tends not to obscure the base. In fact, on small-scale panoramas Berann compiled all foreground and middleground information directly from printed maps, drawing topography in an axonometric fashion that transitions to true perspective deeper in the scene. The dual compilation method, although much easier to execute than a panorama based entirely on true perspective, sometimes appears unrealistic and forced where the axonometric map abruptly changes to the perspective background. The problem is most pronounced on continental panoramas that show the Earth’s curved horizon.
Vertical exaggeration
Panoramas need vertical exaggeration to depict terrain so it approximates our anthropocentric expectations. For example, because an upright human’s vantage point is about 2 meters above the Earth’s surface, even a 100 meter high hill appears significant to us. However, on many small and medium-scale 3D maps, without vertical exaggeration that same 100 meter high hill would barely appear. On Berann’s panoramas vertical exaggeration typically ranges from 1.5:1 to 4:1, depending on the scale and local relief. Small-scale panoramas with low local relief generally display more vertical exaggeration than large-scale panoramas with high local relief.
Berann departed from mapping tradition in his use of selective vertical exaggeration and/or resizing to accentuate important landmarks. (Figure 8) For example, the Denali panorama uses about 2:1 vertical exaggeration throughout the scene, with additional exaggeration applied to the summit of Mt. McKinley. The summit is shown about two times larger in all dimensions (x,y, and z) than the surrounding terrain. By increasing the overall size of Mt. McKinley, Berann avoided the problem of “spiking” that results when too much vertical exaggeration alone is applied to exceptionally tall summits with limited surface area—the Matterhorn would typify this type of mountain.
Carrying the concept of selective vertical exaggeration still further, Berann sometimes varied the vertical exaggeration between elevation zones, depending on the purpose of the panorama and the season of the year. (Figure 9) For example, on a winter ski area map, more vertical exaggeration would be applied to the sloping base (where chair lifts and lodges are located) than to the craggy precipices above. On a summer panorama more exaggeration would be applied to crags to emphasize the scenery that presumably attracts summer visitors (Garfield, 1992).
Berann also liked to emphasize background features in a panorama to show terrain that would normally be too small and distant to comprehend. The Yellowstone panorama is a good example. It shows the greatly enlarged Teton Range along the southern horizon, establishing a familiar geographic context and adding graphical interest to the otherwise flat horizon.
Rotating reality
The Teton Range also illustrates Berann’s very controversial technique of selectively rotating the orientation of mountains and other topographic features within a panorama. Cartographic standards aside, there are compelling reasons for such adjustments. Regardless of how carefully a panoramist chooses the orientation, perspective, and vertical exaggeration, usually a few important landmarks will not appear as clearly as they should. In the southward looking Yellowstone panorama, for example, the north-south trending Teton Range appears as an insignificant nub on the horizon. To make the panorama more meaningful, Berann turned the entire Teton range 55 degrees to show the familiar east face that has appeared in countless photographs. (Figure 10) Because the distant Tetons are used merely as a reference landmark, not unlike a north arrow on a conventional map, their rotation may not be the breach of cartographic ethics it otherwise would seem. Berann wisely applied selective rotation with greater discretion for primary landmarks. For example, in the Denali panorama the summit of Mt. McKinley has been rotated approximately 20 degrees to the east to distinguish between Mt. McKinley’s hard-to-discern North and South Peaks.
Moving mountains
Berann would rearrange and reposition terrain to improve legibility, especially in areas with a concentration of human-made features. One favored technique was to widen narrow mountain valleys notorious for obscuring details within their innermost recesses. Compared to the area shown on a map, Berann widened Yosemite Valley by 220 percent to portray a clearer view of the roads, campgrounds, lodges, and famous landmarks confined to the limited space. Berann also applied a slight straightening to the bends in Yosemite Valley—to look past El Capitan and the other monoliths otherwise obscuring the valley floor. Other terrain movements are done only to improve the graphic composition. In the North Cascades panorama, for instance, the distant Olympic Mountains were dragged about 80 kilometers to the south (left) to align with the upper Skagit River valley, thus creating a visual axis through the center of the scene (see Appendix A for illustration).
Generalization
Considering how much time is spent making a panorama appear realistic, an inevitable question arises: why not just use an oblique aerial photograph instead? The answer is “generalization.” Aerial photographs typically show too much and/or inappropriate detail. How many tourist map sponsors would agree to show unsightly clearcut forests, power lines, or landfills? Oblique aerial photographs also have visibility limits. For example, Berann’s 1986 panorama of Germany depicts an area more than 800 kilometers in length, well beyond the range of low-altitude aerial photographs. (The maximum line of sight ever observed on the Earth’s surface is 370 kilometers.) Moreover, high-altitude aerial photographs and oblique satellite images are inadequate for depicting regions with low relative relief, such as in Germany, without vertical exaggeration.
Panorama generalization is accomplished by manipulating the complexity of the underlying topography and/or surface textures representing vegetation, rocks, etc. Much to Berann’s credit, it is very difficult to detect generalization by comparing his panoramas to topographic maps and 3D computer models—at least for terrain in the foreground and middle ground. In the background areas Berann was selective in the quantity and quality of terrain that is shown. Although one distant mountain range may look like another, Berann went to great efforts to capture their signature characteristics. For example, on the Yellowstone panorama, individual peaks in the far-away Tetons are recognizable despite their stylized depiction and the liberal use of atmospheric haze.
As discussed earlier, Berann sometimes selectively exaggerated the size of important landmarks. This creative license was done at the expense of adjacent or intervening terrain because only a finite space exists on a printed sheet. The sacrificed terrain is usually not omitted but minimized in its extent and functions to bond or connect the more important components of the landscape. On landscapes lacking distinctive topographic features, such as Yellowstone’s Central Plateau, Berann had a particularly difficult time selecting which features to emphasize (Troyer, 2000).
Much of the apparent detail in a Berann panorama derives from the surface texture and detail that is painted on top of the structural landscape itself. The base topography, derived from topographic maps, is often rather generalized. Berann painted subordinate topographic details (microforms), such as the clefts on a cliff face, by referring to oblique aerial photographs. Detail accentuates a scene’s important foreground features and diminishes gradually toward the background. Additional detail applied to well-known landmarks alerts the reader to their importance. Berann lavished attention on the smallest features in a landscape. Cultural features such as roads, dams, and buildings are exaggerated in scale and painted so as to be recognizable. On the Yellowstone panorama, for example, the Old Faithful Lodge, as shown, would be 1.2 kilometers long when compared to a map. Berann also greatly exaggerated the size of waterfalls and Yellowstone’s famous geysers by emphasizing their billowing plumes of mist and steam. (Figure 11)
The intuitive progression of less detail to more detail from background to foreground does not always apply. Careful inspection of the Yellowstone and Denali panoramas reveals that the level of detail increases normally from background to foreground, except in the very closest areas (the bottom edge), which are more generalized. On Denali, foreground generalization is most pronounced in the left and right corners. Foreground generalization tends to direct the viewer’s eye slightly deeper into the scene, where the most important information can be found. It also accentuates perspective by suggesting motion blur, the same effect one would experience by flying into the scene in a high-speed aircraft.
Berann’s NPS panoramas are noteworthy for their apparent lack of generalization on the vertical axis between elevation zones. For the preparation of map shaded relief, Eduard Imhof noted how the aerial perspective effect could enhance three dimensionality by portraying lowlands softer and with slightly less detail than highlands (Imhof, 1982). (Aerial perspective is a graphical technique based on the real-world observation that landscape features farther away from the viewer appear less distinct than those in the foreground.) The aerial-perspective effect is evident on many of Berann’s early panoramas, which show valleys with minimal detail compared to the richly textured mountain peaks above. Later in his career, however, as his artistic skills became more sophisticated and his topographic depictions more realistic, Berann became less reliant on aerial perspective. By the time the NPS panoramas were created he was able to apply equal amounts of detail in lowland and highland areas without compromising three dimensionality.
Many of Berann’s techniques for manipulating landscapes were unorthodox. He often would take questionable liberties (at least from a cartographic perspective) with geographic reality for the convenience of telling a panorama’s story. Whether or not you agree with his techniques, knowing what can be done is valuable for understanding the problems associated with 3D landscape presentation in general.
Perspective
Perspective provides the framework for building a panorama and governs how much of the world the viewer will see. (Figure 6) In general, Berann used more perspective on small-scale scenes than on large-scale scenes. Increasing the amount of perspective increases the field of view, thereby compressing more background information into the finite width of the panorama. Use of perspective enhances realism, unless applied excessively, in which case it tends to pinch background areas unnaturally. Large-scale panoramas employ very little perspective, for the purpose of moving background features closer to the viewer (Garfield, 1992). For example, Berann minimized perspective on the Yosemite panorama to accentuate the lofty peaks along the Sierra Nevada crest, which would be indistinguishable if greater amounts of perspective were used. He compensated for the lack of optical perspective by exaggerating other graphical cues—cloud formations converge toward a false vanishing point, and the differing foreground and background colors enhance aerial perspective. On the North Cascades panorama, on the other hand, Berann used more perspective to pull peripheral landmarks, such as Mt. Rainier, the city of Seattle, and Vancouver Island, into the scene.
Orientation
The cartographic convention of north orientation is not a major factor for determining the orientation (or viewing direction) of large-scale panoramas. None of Berann’s NPS panoramas is oriented due north: North Cascades is oriented to the southwest, Yosemite to the east, Yellowstone to the south, and Denali to the northwest. Berann’s goals when selecting orientation were to maximize the visibility of important local features and to show the broader geographic context. When depicting alpine mountains in the northern hemisphere, such as the Alps or North Cascades, he often used southwest orientation. Looking southwest reveals northeast mountain faces, which tend to be steeper, more distinctive, and more glaciated. Prevailing southwest winds transport summit snows to these lee slopes, which also get less direct sunlight.
Berann preferred the view from lowlands toward highlands. In fact, most of his panoramas have a lofty mountain range as the backdrop. On large scale panoramas, viewers tend to become disoriented when looking from highlands toward lowlands, despite our real-world familiarity with downhill vistas from mountain peaks. This phenomenon received prominent attention during the controversial court martial of the U.S. pilot who clipped the cable of a ski gondola in the Italian Alps in 1998, tragically killing 20 people. The legal defense team for the pilot presented a digital fly-through animation, which was shown on CBS national television news in the U.S., to demonstrate the difficulties of judging elevation when traveling downhill through a narrow mountain valley (Visual Forensics, 1999).
Berann sometimes added twist (skewed orientation) to his scenes to show topographic features with cultural importance that otherwise would be outside the field of view. This is evident in two NPS panoramas: Yosemite was skewed to the north to include Hetch Hetchy Reservoir, the site of a landmark conservation-versus-development battle in the early nineteen hundreds. North Cascades was skewed to the south to include Mt. Rainier—the defining landmark for the millions of people who live in Seattle and the Puget Sound lowlands (see Appendix A for illustration).
Projection plane
Cartographically speaking, Berann did not accept that the Earth is flat, even on large-scale panoramas. He added curvature to the projection plane (the theoretical flat base upon which 3D terrain projects upwards) to enhance viewing. On a typical 3D scene with a low-elevation view, the horizon is visible, and the landscape looks realistic, but tall features in the foreground obscure the background. Conversely, high elevation views show background terrain better, but, without the horizon, they look too much like conventional maps.
To solve this problem Berann emulated the view as seen from an airplane. (Figure 7) From high above the Earth the horizon is always visible, yet when you shift your eyes downward the view gradually becomes less oblique and more planimetric. To bring this effect to a panorama, Berann tilted the projection plane toward the viewer and, from a point about two thirds of the way into the scene, added convex curvature to flatten the horizon. The end result is a panorama that combines the best of both worlds: the foreground and middleground (where the important information resides) appear map-like while the background appears realistic, complete with a horizon and sky (Patterson, 1999).
Berann’s manipulation of the projection plane is most evident on small scale panoramas, such as his 1986 view of Germany, in which the relatively low relief tends not to obscure the base. In fact, on small-scale panoramas Berann compiled all foreground and middleground information directly from printed maps, drawing topography in an axonometric fashion that transitions to true perspective deeper in the scene. The dual compilation method, although much easier to execute than a panorama based entirely on true perspective, sometimes appears unrealistic and forced where the axonometric map abruptly changes to the perspective background. The problem is most pronounced on continental panoramas that show the Earth’s curved horizon.
Vertical exaggeration
Panoramas need vertical exaggeration to depict terrain so it approximates our anthropocentric expectations. For example, because an upright human’s vantage point is about 2 meters above the Earth’s surface, even a 100 meter high hill appears significant to us. However, on many small and medium-scale 3D maps, without vertical exaggeration that same 100 meter high hill would barely appear. On Berann’s panoramas vertical exaggeration typically ranges from 1.5:1 to 4:1, depending on the scale and local relief. Small-scale panoramas with low local relief generally display more vertical exaggeration than large-scale panoramas with high local relief.
Berann departed from mapping tradition in his use of selective vertical exaggeration and/or resizing to accentuate important landmarks. (Figure 8) For example, the Denali panorama uses about 2:1 vertical exaggeration throughout the scene, with additional exaggeration applied to the summit of Mt. McKinley. The summit is shown about two times larger in all dimensions (x,y, and z) than the surrounding terrain. By increasing the overall size of Mt. McKinley, Berann avoided the problem of “spiking” that results when too much vertical exaggeration alone is applied to exceptionally tall summits with limited surface area—the Matterhorn would typify this type of mountain.
Carrying the concept of selective vertical exaggeration still further, Berann sometimes varied the vertical exaggeration between elevation zones, depending on the purpose of the panorama and the season of the year. (Figure 9) For example, on a winter ski area map, more vertical exaggeration would be applied to the sloping base (where chair lifts and lodges are located) than to the craggy precipices above. On a summer panorama more exaggeration would be applied to crags to emphasize the scenery that presumably attracts summer visitors (Garfield, 1992).
Berann also liked to emphasize background features in a panorama to show terrain that would normally be too small and distant to comprehend. The Yellowstone panorama is a good example. It shows the greatly enlarged Teton Range along the southern horizon, establishing a familiar geographic context and adding graphical interest to the otherwise flat horizon.
Rotating reality
The Teton Range also illustrates Berann’s very controversial technique of selectively rotating the orientation of mountains and other topographic features within a panorama. Cartographic standards aside, there are compelling reasons for such adjustments. Regardless of how carefully a panoramist chooses the orientation, perspective, and vertical exaggeration, usually a few important landmarks will not appear as clearly as they should. In the southward looking Yellowstone panorama, for example, the north-south trending Teton Range appears as an insignificant nub on the horizon. To make the panorama more meaningful, Berann turned the entire Teton range 55 degrees to show the familiar east face that has appeared in countless photographs. (Figure 10) Because the distant Tetons are used merely as a reference landmark, not unlike a north arrow on a conventional map, their rotation may not be the breach of cartographic ethics it otherwise would seem. Berann wisely applied selective rotation with greater discretion for primary landmarks. For example, in the Denali panorama the summit of Mt. McKinley has been rotated approximately 20 degrees to the east to distinguish between Mt. McKinley’s hard-to-discern North and South Peaks.
Moving mountains
Berann would rearrange and reposition terrain to improve legibility, especially in areas with a concentration of human-made features. One favored technique was to widen narrow mountain valleys notorious for obscuring details within their innermost recesses. Compared to the area shown on a map, Berann widened Yosemite Valley by 220 percent to portray a clearer view of the roads, campgrounds, lodges, and famous landmarks confined to the limited space. Berann also applied a slight straightening to the bends in Yosemite Valley—to look past El Capitan and the other monoliths otherwise obscuring the valley floor. Other terrain movements are done only to improve the graphic composition. In the North Cascades panorama, for instance, the distant Olympic Mountains were dragged about 80 kilometers to the south (left) to align with the upper Skagit River valley, thus creating a visual axis through the center of the scene (see Appendix A for illustration).
Generalization
Considering how much time is spent making a panorama appear realistic, an inevitable question arises: why not just use an oblique aerial photograph instead? The answer is “generalization.” Aerial photographs typically show too much and/or inappropriate detail. How many tourist map sponsors would agree to show unsightly clearcut forests, power lines, or landfills? Oblique aerial photographs also have visibility limits. For example, Berann’s 1986 panorama of Germany depicts an area more than 800 kilometers in length, well beyond the range of low-altitude aerial photographs. (The maximum line of sight ever observed on the Earth’s surface is 370 kilometers.) Moreover, high-altitude aerial photographs and oblique satellite images are inadequate for depicting regions with low relative relief, such as in Germany, without vertical exaggeration.
Panorama generalization is accomplished by manipulating the complexity of the underlying topography and/or surface textures representing vegetation, rocks, etc. Much to Berann’s credit, it is very difficult to detect generalization by comparing his panoramas to topographic maps and 3D computer models—at least for terrain in the foreground and middle ground. In the background areas Berann was selective in the quantity and quality of terrain that is shown. Although one distant mountain range may look like another, Berann went to great efforts to capture their signature characteristics. For example, on the Yellowstone panorama, individual peaks in the far-away Tetons are recognizable despite their stylized depiction and the liberal use of atmospheric haze.
As discussed earlier, Berann sometimes selectively exaggerated the size of important landmarks. This creative license was done at the expense of adjacent or intervening terrain because only a finite space exists on a printed sheet. The sacrificed terrain is usually not omitted but minimized in its extent and functions to bond or connect the more important components of the landscape. On landscapes lacking distinctive topographic features, such as Yellowstone’s Central Plateau, Berann had a particularly difficult time selecting which features to emphasize (Troyer, 2000).
Much of the apparent detail in a Berann panorama derives from the surface texture and detail that is painted on top of the structural landscape itself. The base topography, derived from topographic maps, is often rather generalized. Berann painted subordinate topographic details (microforms), such as the clefts on a cliff face, by referring to oblique aerial photographs. Detail accentuates a scene’s important foreground features and diminishes gradually toward the background. Additional detail applied to well-known landmarks alerts the reader to their importance. Berann lavished attention on the smallest features in a landscape. Cultural features such as roads, dams, and buildings are exaggerated in scale and painted so as to be recognizable. On the Yellowstone panorama, for example, the Old Faithful Lodge, as shown, would be 1.2 kilometers long when compared to a map. Berann also greatly exaggerated the size of waterfalls and Yellowstone’s famous geysers by emphasizing their billowing plumes of mist and steam. (Figure 11)
The intuitive progression of less detail to more detail from background to foreground does not always apply. Careful inspection of the Yellowstone and Denali panoramas reveals that the level of detail increases normally from background to foreground, except in the very closest areas (the bottom edge), which are more generalized. On Denali, foreground generalization is most pronounced in the left and right corners. Foreground generalization tends to direct the viewer’s eye slightly deeper into the scene, where the most important information can be found. It also accentuates perspective by suggesting motion blur, the same effect one would experience by flying into the scene in a high-speed aircraft.
Berann’s NPS panoramas are noteworthy for their apparent lack of generalization on the vertical axis between elevation zones. For the preparation of map shaded relief, Eduard Imhof noted how the aerial perspective effect could enhance three dimensionality by portraying lowlands softer and with slightly less detail than highlands (Imhof, 1982). (Aerial perspective is a graphical technique based on the real-world observation that landscape features farther away from the viewer appear less distinct than those in the foreground.) The aerial-perspective effect is evident on many of Berann’s early panoramas, which show valleys with minimal detail compared to the richly textured mountain peaks above. Later in his career, however, as his artistic skills became more sophisticated and his topographic depictions more realistic, Berann became less reliant on aerial perspective. By the time the NPS panoramas were created he was able to apply equal amounts of detail in lowland and highland areas without compromising three dimensionality.
